Today's AI Pulse
-1.png)
Anthropic shuts down Fable 5 and Mythos 5 models under U.S. export controls
Anthropic abruptly disabled its flagship Fable 5 and Mythos 5 models to comply with a U.S. export‑control directive, underscoring the vulnerability of firms that depend on closed‑source AI services. The move triggered a sharp rally in Chinese open‑source offerings as investors shifted to downloadable models that can run on‑premise.
Also developing:
By the numbers: Respond raises $62.5M Series B
Broadcom Q2 FY26 Revenue Beats Forecast, EPS Miss Fuels AI Revenue Debate
Broadcom reported $22.2 billion in Q2 FY26 revenue, a 48% year‑on‑year jump driven by AI semiconductors, yet earnings per share fell short of analysts' whisper forecasts. The mixed results triggered a sell‑off in Asian and European tech stocks as investors weighed the sustainability of the AI boom against a softer earnings outlook.
Thrad Launches Four Native Ad Formats for LLM & AI Conversation Environments
Thrad, an AI‑native advertising infrastructure platform, announced the general availability of four purpose‑built ad formats for large language model (LLM) and conversational AI environments. The formats—Card, Prompt, Carousel and Poll—appear inline with AI responses and use CPM, CPC or cost‑per‑engagement...
Ofcom Publishes Latest Approach to Artificial Intelligence
Ofcom released its third annual AI strategic approach for 2026/27, outlining technology‑neutral, outcomes‑based regulatory principles. The report includes a broadcast sector case study that maps AI use in content creation and generative‑AI value chains, while identifying barriers such as copyright...
AI Can Now Coach Amateur Virologists, and Top Tech Leaders Want Congress to Act on DNA Security
In an open letter, leading technology CEOs and prominent scientists urged Congress to mandate federal screening of all synthetic DNA orders. They highlight that AI tools now answer virology lab‑procedure questions better than PhD‑level experts, lowering barriers for malicious actors....
Freshflow Lands $10 Million Series A to Scale AI Platform Cutting Grocery Waste
Freshflow, the Berlin‑based AI food‑waste startup, closed a $10 million Series A round led by Reimann Investors. The funding will fund expansion of its forecasting platform beyond produce into meat, bakery and other perishable categories, and accelerate rollout across Europe. Investors cite...
Netskope Q1 ARR Hits $845M, Up 29% YoY on AI Security Surge
Netskope announced Q1 2027 ARR of $845 million, a 29% year‑over‑year rise, driven by strong product adoption and a 113% net retention rate. The cloud‑security SaaS firm also lifted its full‑year revenue guidance and unveiled five AI‑focused security products.
Inside IVA Studio: Architecting the Autonomous Procurement Workforce
Ivalua unveiled IVA Studio at its NOW 2026 conference in Paris, a no‑code platform that lets procurement teams create, orchestrate and govern autonomous AI agents—called skills—directly on their Source‑to‑Pay data. The solution ships with a library of ready‑to‑deploy skills organized into...
HPE Posts Record $10.7B Quarter on AI Server Boom, Shares Jump Over 20%
Hewlett Packard Enterprise announced fiscal Q2 revenue of $10.7 billion, up 40% year‑over‑year, as AI‑optimized servers and networking gear fueled a surge in sales. The results sent HPE stock soaring more than 20% and prompted the company to lift its full‑year...
Scarcity Is Driving AI Innovation Outside Silicon Valley
Rising compute costs and energy constraints are prompting AI infrastructure development outside traditional hubs. India’s Yotta Data Services, Africa’s Cassava Technologies, Brazil’s sovereign AI factory, and the UAE’s Core42 are building large GPU clusters powered by renewable or local sources....
Machine Learning System Design Interview #47 - The EWC Rigidity Trap
Elastic Weight Consolidation (EWC) can prevent catastrophic forgetting but may over‑regularize, causing the model to become too rigid for new tasks. In a DeepMind interview scenario, the EWC penalty dominates the loss landscape, leading to a capacity lockout where essential...
Frontline Workers Twice as Likely to Use Unapproved AI
Mitel’s State of Workforce Communication report finds frontline workers are twice as likely to turn to unapproved AI tools, highlighting a widening gap between AI investment and practical enablement. While 93% of IT leaders deem communication platforms critical, only 34%...
Mint Explainer: What Do the Supreme Court’s Draft AI Rules Mean for India's Justice System?
On 3 June 2026 the Supreme Court of India issued draft regulations to govern the use of artificial intelligence in its judicial system, opening a public comment window until 20 June. The proposals allow AI tools to assist judges and court staff but...
Raising the Bar: The Non-Negotiable Requirements for Biometric Identity in Air Travel
Airports are shifting biometric identity from pilots to operational expectations, with IATA reporting 20‑50% faster passenger processing and the ability to reallocate 10‑15% of checkpoint staff to higher‑value duties. Yet real‑world deployments deliver mixed results because of lighting variability, demographic...
Microsoft's New Models Give It a Better Moat
Microsoft unveiled a suite of seven new in‑house AI models at Build, highlighted by the flagship MAI‑Thinking‑1—a 35‑billion‑parameter mixture‑of‑experts model with a 256k‑token window and built on clean, licensed data. The company positioned the model as a direct competitor to...
AI Is Ushering in a New Era of Colonialism
Artificial intelligence is increasingly built on data harvested from Western sources, prompting critics to label the practice a new form of colonialism. Large language models rely on text generated by predominantly white, male authors, reproducing those cultural biases in their...
The Next AI Breakthrough Won’t Come From Bigger Models, but From Better Data
The article argues that the next AI breakthrough will stem from better, domain‑specific data rather than ever larger models or faster chips. It highlights a growing "data gap" where sectors like healthcare, enterprise workflows, and multilingual speech lack the structured,...
Interview with CreateMe CEO Campbell Myers: From Stitching to Bonding – Physical AI Could Transform the Way Clothes Are Made
CreateMe, led by CEO Campbell Myers, is redesigning garment manufacturing by swapping traditional stitching for adhesive‑based assembly powered by robotics and physical AI. The platform combines intelligent grippers, real‑time fabric state sensing, and thermoreversible adhesives to handle the unpredictable behavior...
Singapore Advances National AI Strategy to Drive Sector-Wide Transformation
Singapore is expanding its National AI Strategy to move AI from pilots to sector‑wide deployment across connectivity, advanced manufacturing, healthcare and finance, which together account for over 40% of GDP. The plan, outlined by Minister Josephine Teo, emphasizes building compute...
Context Engineering Gains Spotlight in BMJ Digital Health
I've been talking about the importance of context engineering for a year (even in my new digital course). Besides prompt engineering, the ability to work efficiently with large language models, context engineering goes one step further. It selects and structures...
Indeed Says It's Time to Stop Searching and Start Matching
Indeed is rebranding as an AI‑powered hiring platform with the global "Jobs need people" campaign, developed with 72andSunny. The rollout spans TV, streaming, YouTube, social and sports partnerships, including a "Chief World Cup watcher" activation with FOX Sports. Indeed now...
Asana Launches AI-Powered Products to Help Organizations Manage Human and Agent Work
Asana unveiled an AI‑powered suite called Agentic Work Management at its Work Innovation Summit in London. The suite includes the Dash AI agent, a chief‑of‑AI staff for each user, and upgraded AI teammates that integrate with tools like Gmail, Slack...
Lundbeck Partners with Cradle to Accelerate AI-Driven Biotherapeutics Discovery
Lundbeck has teamed up with AI‑driven biotech Cradle to apply generative protein‑engineering technology to its drug pipeline. The collaboration will initially target two antibody programs aimed at central nervous system disorders, using iterative experimental feedback to accelerate candidate discovery and...
Verified Machine Learning Infrastructure
A 2026 survey of 23 experts across formal methods, AI infrastructure, cybersecurity, and policy examined whether mathematical verification techniques can mitigate risks in frontier AI systems. Respondents identified inference runtimes, model compilers, and training pipelines as the most verification‑friendly components,...
XAI Updates Grok Imagine to 1.5 with Image-to-Video Generation at 720p Resolution
Elon Musk’s xAI unveiled Grok Imagine Video 1.5, an image‑to‑video model that transforms a single still into a short clip up to 720p. Users can dictate camera moves, pacing and mood via text prompts, while the system preserves the original lighting...
The Invisible Shopper Rewriting Asia’s E-Commerce Playbook
The Morph "Agentic Economy" report reveals that AI‑driven software buyers are already reshaping retail, with $67 billion of Cyber Week spend in 2025 directly influenced by agents. Open standards such as Google’s AP2, Stripe‑OpenAI’s Agentic Commerce Protocol, and Ethereum’s ERC‑8004 enable autonomous...
This Startup Is Trying to Make Peace Between AI Companies and Creatives
A startup called Created by Humans, founded by former Scribd CEO Trip Adler, is building a licensing platform that lets AI developers use books and other creative works only with explicit permission from rights holders. The move comes after high‑profile...
Foxconn, Intel Team Up to Develop AI Infrastructure
Taiwan’s Foxconn and Intel announced a partnership to build next‑generation artificial‑intelligence infrastructure for modern data centers. The collaboration will focus on alleviating inference and agentic workload bottlenecks by co‑designing custom AI chips and integrating Foxconn’s high‑volume manufacturing with Intel’s silicon...
UN Recommends Omitting Politeness With AI To Save Power
A United Nations University report finds that dropping polite words like “please” and “thank you” when chatting with AI can slash token output by roughly 30%, translating into 87‑98 GWh of electricity saved each year. The savings equal the residential power...
Nedbank, Jumo Bet on AI Lending for the Underbanked
Nedbank has teamed with fintech Jumo to launch an AI‑driven short‑term loan product, Nedbank Quick Loans, inside its Money App. The digital offering provides funds ranging from R500 ($27) to R50,000 ($2,700) in as little as five minutes, with no...
Rethinking Infrastructure Access in the Age of Agentic AI
The rise of agentic AI is reshaping identity and access management, exposing gaps in traditional static credential models. AI agents require dynamic, just‑in‑time privileges and real‑time session monitoring to prevent unauthorized actions. HashiCorp Boundary, paired with Vault, delivers unique agent...
The Edge LLM Offload Story
Developers seeking on‑device large language models now have a dedicated solution as Synaptics and Google Research combine the Astra SL2610 processor with the Coral NPU to form the Torq NPU. The platform tackles edge‑LLM bottlenecks through static model conversion, hardware‑accelerated...
How Vanguard Is Preparing for an AI-Powered Future
In this Technovation episode, Vanguard’s CIO Nitin Tandon, personal‑wealth head Joanna Ruttenberg, and chief economist Joe Davis discuss how the firm is embedding AI across its operations to boost agility, personalization, and client outcomes. They trace Vanguard’s multi‑year modernization—moving 90%...
AI and the New Value Equation - the Squiggly Line Is the Journey (3/3)
The article challenges traditional linear AI maturity models, proposing a wavy “Hands‑On → Hands‑Ready → Hands‑Off” journey that reflects the iterative reality of deploying autonomous agents. Hands‑On describes human‑led firms using AI tools, while Hands‑Ready captures the turbulent phase of...
The Channel Is Heading Straight for an AI Infrastructure Wall
Channel partners have excelled at launching AI pilots, but moving those proofs‑of‑concept into production exposes a severe infrastructure gap. Legacy storage, networking and power constraints cannot sustain the high‑throughput demands of generative AI, and 80% of the problem lies in...
Kodesage Raises $6.6M for AI-Powered Legacy Software Modernisation
Kodesage, a 2024 startup, raised $6.6 million in a seed round led by VentureFriends to accelerate its AI‑powered platform for legacy software modernization. The on‑premise solution extracts code and documentation from aging stacks such as COBOL, Oracle Forms, PL/SQL, PowerBuilder...
The AI Production Paradox: Why Customer Agents Are Becoming a Brand Risk for Marketers
Sinch’s AI Production Paradox report finds that 62% of enterprises already run customer‑facing AI agents in production, and 74% have rolled back or shut down an agent after launch because of governance failures. The problem is most acute for marketers,...
The Future of “AI Citizenship”: Sovereign AI as the New Economic Currency
The article introduces “AI citizenship,” a vision where nations grant digital residency and access to sovereign AI infrastructure as a strategic economic incentive. Governments across the Middle East, Asia and Europe are building national foundation models, sovereign clouds, and compute...
Revolutionizing Deep Space Exploration with AI and Machine Learning
Artificial intelligence and machine learning are reshaping deep‑space missions, with the AI‑in‑space market projected to rise from $5.9 billion in 2025 to $7.8 billion in 2026, a 32.3% CAGR. NASA has embedded AI into its Mars rovers—Curiosity and Perseverance—enabling autonomous navigation, target...
The Hidden Bottleneck in LLM Inference and the Impact on MLPerf Benchmarking
Recent LLM inference benchmarks reveal that GPUs excel at the prefill phase but falter when real‑time token generation is required. The generation step is sequential and memory‑bound, causing utilization to collapse as batch sizes shrink for low‑latency serving. Disaggregating workloads—dedicating...
Neurosymbolic AI Will Outperform LLMs, Render Big Bets Premature
Three Predictions: 1. Some form of AI, probably neurosymbolic in nature, will come that is far more economical and data- and energy-efficient than LLMs, and it will make an absolute fortune. 2. LLMs, on the other hand, will never be all...
Huawei's Tau Scaling Law Reveals Hidden Inference Bottleneck
Most AI teams are optimizing the model. But the real bottleneck in inference is underneath it. Huawei's Tau Scaling Law (Her's Law) was just introduced at IEEE ISCAS in Shanghai. It reframes how we think about AI performance entirely. Here's the breakdown… #HuaweiPartner @huawei
AI Infrastructure Company Jentic Wants to Score the APIs AI Agents Will Depend On
Jentic, an AI infrastructure firm based in Dublin, has launched a free API Scoring tool that measures how ready enterprise APIs are for AI agents. First released in December 2025, the tool evaluates APIs across six dimensions—technical correctness, clarity, behavioral...
NVIDIA AI
NVIDIA-Powered #AI Farming Machine Uses Precision Lasers to Eliminate Weeds Without Chemicals via @Rainmaker1973 #AgriTech #Innovation #Technology #TechForGood https://t.co/1YyqXfY1vd
China Unveils National Roadmap to Strengthen AI Metrology Capabilities
China’s State Administration for Market Regulation and the National Development and Reform Commission released a national roadmap to build AI metrology capabilities. The plan outlines six pillars—foundational support, general and core technologies, standards, service industry, and intelligent empowerment—to create a...
AI Boom Creates Connectivity Challenge for Integrators
Altnets’ latest whitepaper warns that the AI boom is turning connectivity into the next bottleneck for system integrators. Rapid AI adoption is driving a six‑fold increase in global datacentre capacity by 2035, pushing demand for dense fibre, backhaul and edge...
How to Use AI to Grow Your Online Store Sales: 4 Tools
Steve Chou, who runs the 7‑figure ecommerce site BumblebeeLinens.com, leverages AI in four revenue‑focused ways: an AI‑powered on‑site search that eliminated null queries and quadrupled search‑driven sales, AI‑driven cross‑sell recommendations that lifted average order value 22%, AI analysis that surfaces...
My Year with the Robots: How Joanna Stern Let AI Into Her Home, Work – and Heart
Joanna Stern, former Wall Street Journal tech columnist, spent 2025 living with artificial intelligence in every facet of her home and work, from answering texts and cooking meals to driving her car and analyzing mammograms. The year‑long experiment culminated in...
AI Saves Workers a Day a Week, but They Don’t Know What to Do with It
Boston Consulting Group’s Global AI at Work Survey finds 42% of frontline employees save a full day each week thanks to AI, yet 66% lack guidance on how to redeploy that time. AI usage has surged, with 74% of workers...
What AI Agents Actually Mean for Enterprise Knowledge Work
Enterprise buying is increasingly fragmented, with multi‑stakeholder committees shaping decisions before vendors engage. AI agents are emerging as a layer that assembles, contextualizes, and delivers precise knowledge to each participant, reducing search time and improving answer relevance. The technology’s impact...
AI Agents as Customers Blur Legal Liability
On @DEXintheCityPod, @kkirkbos and Jessi Brooks on what happens when your AI agent becomes the customer — and the law has no idea who's liable if something goes wrong. New clip: 🎧 https://t.co/8wmRNy8W5u